A recent paper by the CDMS collaboration (PRL here and free text here) makes this statement in the abstract:

This blind analysis of 140.2 kg day of data taken between July 2007 and September 2008 revealed three WIMP-candidate events...The probability that the known backgrounds would produce three or more events in the signal region is 5.4%.

I am not part of the particle physics communities or the dark matter communities so my questions are:

  1. Is this convincing evidence for the existence of WIMPs?
  2. What does this say, if anything, about the overall composition of dark matter? I.E. is this convincing evidence that a large fraction of dark matter is from WIMPs?
  • $\begingroup$ 1) arXiv:1306.3983, 1306.5534 No dark matter <BR> 2) arXiv:1310.4009 Milgrom acceleration can be empirically sourced. arXiv:1306.3983 = arxiv.org/abs/1306.3983 $\endgroup$
    – Uncle Al
    Feb 18, 2014 at 17:25

2 Answers 2


The excesses have looked convincing to many people but they don't look convincing anymore. Last October, LUX in South Dakota presented the results of their superior analysis


which safely excluded the theories of WIMP dark matter directly suggested by CDMS II Si and other experiments. The complete and pure absence of a signal in LUX shows almost certainly that the excesses in CDMS II Si and other experiments were due to some overlooked background (non-dark-matter-related) processes.

WIMP is still plausible and attractive as a model of dark matter. But there's no evidence one way or another which is why physicists keep on studying it and other models as well, including e.g. those of the sterile neutrino dark matter or axions.


I don't know the paper you cite, but you quote a $p$-value of $5.4 \%$ which is not conclusive at all. Given that there are many experiments searching for dark matter, a fluctuation like this is extremely likely to happen by pure chance. This is known as look-elsewhere effect. Even without considering this effect, a $p$-value of $5.4 \%$ is nothing to get excited about.

In order to not proclaim a discovered particle every now and then and subsequent realization that it was just a statistical fluctuation, tight requirements for a particle discovery have been set:

To claim "observation" of a new particle, a $3 \sigma$ effect needs to be observed. The chance that this happens by a pure statistical fluctuation is $0.3 \%$. "Discovery" of a new particle is claimed when a $5 \sigma$ effect is observed in data. The probability of this being just due to a fluctuation is $3 \cdot 10^{-7}$.

Note that different standards exist, the agreement of $3 \sigma$ and $5 \sigma$ is arbitrary and it does not protect you from potential mistakes in your analysis. For example, the superluminal neutrinos were reported to be a $\sim 6 \sigma$ effect, but hardly anyone believed it, not because of statistical fluctuation, but because of systematic errors.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.